Ecol Res (****) **: *** ***

DOI **.****/s*****-**5-0065-6

O R I GI N A L A R T IC L E

Hui Zeng Daniel Z. Sui X. Ben Wu

Human disturbances on landscapes in protected areas:

a case study of the Wolong Nature Reserve

Received: 21 October 2004 / Accepted: 9 February 2005 / Published online: 13 April 2005

The Ecological Society of Japan 2005

protected areas. Future management of Wolong Nature

Abstract Human-induced ecological degradation in

Reserve should focus on adoption of e ective policies to

protected areas is of great concern in landscape ecolo-

further constrain human activities in the reserve.

gical studies. Using Landsat TM data and GIS-based

spatial analysis, we assessed the impacts of human dis-

turbances on landscape structure in the Wolong Keywords Human disturbances Landscape

Nature Reserve in southwestern China. Bu er zone and dissimilarity Landscape structure Wolong Natural

landscape dissimilarity analysis were used to examine Reserve

the scope of three types of human-induced

disturbances construction of hydropower stations,

human activities around settlement, and human activi-

ties along roads. We found that the impacts of these Introduction

human disturbances extend to a threshold distance of

about 1,000 m from the sources of disturbance. The Human-induced disturbances on landscape are sub-

intensity of the impact of human disturbances on land- stantial and growing, as documented by an extensive

scape structure exhibited clear distance decay e ects. and growing interdisciplinary literature (Turner et al.

The rst 200 m bu er zone is the area where human 1990; Goudie 2000). Vitousek et al. (1997) estimated that

activities have in icted the most visible changes, with a human activities had altered approximately one-third to

decrease of forest cover by 15 40% and an increase of one-half of the Earth s land surface in many signi cant

shrub and barren land area by 15 50%. The relative ways. Indeed, we live on an increasingly human-domi-

intensity of the overall impact on landscape structure nated planet, and human activities had generated enor-

was highest around hydropower stations, second around mous impacts on all the four spheres of planet

human settlements, and lowest along roads. Overtime, earth atmosphere, hydrosphere, lithosphere, and bio-

however, the relative level of impact associated with sphere. Accelerating human impacts on the environment

construction of hydropower stations will likely decrease have led to dramatic, in some cases even irreversible,

and that associated with human activities around set- changes of the environmental dynamics from local to

tlements likely increase. Our case study of Wolong ex- global scales (Pimm and Lawton 1998). With the

poses hurdles that habitat preservation must cross in momentum of human population growth and mandates

for further economic development in both developing

and developed countries, human impacts on the envi-

H. Zeng

ronment will most likely continue, and consequently,

School of Environmental Science,

further deteriorate already fragile ecosystems (Ehrlich

Peking University, Beijing, 100871, P.R. China

1995; Vitousek et al. 1997). Not surprisingly, how to

D. Z. Sui reduce human disturbances on landscape is one of the

Department of Geography,

major challenges humanity faces as we enter the twenty-

Texas A&M University,

College Station, TX 77843-3147, USA rst century.

E-mail: abpnvr@r.postjobfree.com

As a major conservation policy to mitigate these

Tel.: +1-979-***-****

continuing trends toward ecological degradation, nature

Fax: +1-979-***-****

reserves have been designated worldwide to protect

X. B. Wu

endangered species and ecosystems (Wright 1996; IUCN

Department of Rangeland Ecology and Management,

1998). By 1996, more than 12,700 protected areas were

Texas A&M University,

established around the world 13.2 million km2 in total

College Station, TX 77843, USA

488

coverage, or approximately 8.8% of the Earth s land

Methods

surface (Dompka 1996). These protected areas consti-

tute a cornerstone for conserving biodiversity. Despite

Study area

sustained debates about the e ectiveness of protected

areas for conservation (Ghimire and Pimbert 1997;

Our study area is the Wolong Nature Reserve a ag-

Brandon et al. 1998; Brooks et al. 2001; Bruner et al.

ship protected area established in 1975 for the protection

2001; Vanclay et al. 2001), the majority of the literature

of the endangered giant pandas (Ailuropoda melanoleu-

has con rmed the necessity and value of establishing

ca) (Schaller et al. 1985). Located in Sichuan Province,

such nature reserves for biodiversity conservation

(102-**-***-** E, 30 45

southwestern China

(McNeely et al. 1995; Forester and Machlis 1996;

31 25 N), Wolong Natural Reserve is so far the largest

Hocking et al. 2000; Bruner et al. 2001).

nature reserve designated speci cally to protect and

Research from a large number of protected areas

conserve the endangered giant pandas.

indicates more success than failure in protecting habitats

Geographically, Wolong Nature Reserve is situated

in the face of serious environmental threats (Maikhuri

on the south side of Qionglai Mountain on the east side

et al. 2000; Samant et al. 2000; Trakolis 2001). Generally

of the Hengduan Mountain Range. Wolong Natural

speaking, biodiversity can potentially be better protected

Reserve is located in the upper reaches of the Minjiang

from the impacts of human activities after an area has

River and within the jurisdiction of Wenchuan County,

been designated as a nature reserve, although

Aba Zang Autonomous State in Sichuan Province. Its

encroachments and threats are still common in many

total area is 2,000 km2, 52 km from east to west and

protected areas (McNeely et al. 1995; Dompka 1996).

62 km from south to north (Fig. 1). Since the beginning

The study by Homewood et al. (1998) revealed that

of the twentieth century, this area has been impacted by

population growth alone is a necessary but not a su -

large-scale logging activities. Beginning in 1965, it served

cient condition to initiate changes in landscape struc-

as the timber production area for Hongqi Forestry Bu-

ture. It is often the various human activities (such as the

reau of Sichuan Province, a state-owned forestry man-

expansion of road networks or the construction of

agement company, until 1975 when Wolong Natural

industrial/living facilities, etc.) that will ultimately lead

Reserve was established with rati cation by the State

to major landscape structural changes, and yet the scope

Council. Since its establishment, commercial logging

and nature of human activities in the protected areas

activities have been prohibited. In 1980, as proposed by

have not been well understood. Although it is well

China s National Committee for Man and Biosphere

known that excessive human activities can cause irre-

and approved by UNESCO and the UN Executive Bu-

versible environmental degradation, sealing a natural

reau for Man and Biosphere, Wolong Nature Reserve

area o from the outside human world is not necessarily

joined the World Natural Reserves Network, whose

the best way to protect the area. On the contrary, such a

main protection targets are alpine ecosystems and rare

policy may sooner or later destroy the area it was in-

animal species like giant pandas (Qin and Allen 1993).

tended to protect (Primack 2000). In other words, good

As part of a network of 33 protected areas to protect

intentions do not always bring about desirable conse-

pandas, it provides habitat for approximately 10% of

quences and, in the long run, according to the revenge

the wild panda population and is the world s most

theory, nature tends to get even with us (Tenner 1996).

widely recognized conservation icon (MacKinnon and

Despite several decades of research on nature reserves

DeWulf 1994; Zhang and Hu 2000). As a national

and biodiversity, we still have limited knowledge about

agship nature reserve with international recognition,

how human activities have impacted landscape in pro-

Wolong Natural Reserve has received steady nancial

tected areas, and what constitutes the appropriate level

and technical assistance from the Chinese government as

of human activities in protected areas.

well as various international organizations (James et al.

Our research seeks to make a modest step toward

1999). The Wolong Natural Reserve research station is

understanding the e ects of human disturbances on

one of the best-funded and sta ed facilities for the

landscape structure in protected areas through a case

captive breeding of giant pandas (Zhang et al. 2000). As

study of the Wolong Nature Reserve. Speci cally, we

in many protected areas, people live in Wolong, with

want to answer the following two questions: (1) How do

approximately 4,260 local residents and 904 households

various types of human disturbances manifest their im-

inside the reserve in 1995, compared to 2,560 people in

pacts di erentially over landscapes in scale and inten-

421 households in 1975. Human activities include, but

sity? and (2) How do landscape structures a ected by

are not limited to, agriculture, fuel-wood collection,

human disturbances di er from those in areas free from

timber harvesting, road construction and maintenance,

human disturbances? Answering such questions is cru-

Chinese herbal medicine collection, and tourism (Lu

cial to developing e ective policies for better manage-

et al. 2003).

ment of the Wolong Nature Reserve and protected areas

Due to strict conservation and management policies,

elsewhere. This study will contribute to our under-

Wolong Nature Reserve is now free of the most exten-

standing of the scale and scope of human disturbances

sive human disturbances (such as large-scale logging

and their impacts on landscape structure in protected

activities) of its natural ecosystems. Human impacts are

areas.

489

Fig. 1 The location of Wolong

Natural Reserve and the

sources of human disturbances

con ned primarily to limited areas (Swaisgood et al. classi cation method in IDRISI was used to classify

2001). We have identi ed three main sources of human the image based on raw bands and NDVI to generate

impacts in our study in the recent past: road construc- the vegetation map. Six vegetation types were used in

tion and maintenance, human settlement areas, and the the classi cation: coniferous forest, mixed coniferous

construction/operation of small hydropower stations. broadleaf forest, broadleaf forest, shrub, herbaceous,

There are currently 10 hydropower stations, 47 settle- and barren land. A total of 43 sites (10 coniferous

ments, and 496 km of roads in Wolong Natural Reserve forest, 11 mixed forest, 12 broadleaf forest, 6 shrub,

(Fig. 1). In the mid-1980s, the local government built and 4 herbaceous sites) were sampled and used as

several small hydropower stations in order to supply training sites for classi cation, along with another 15

energy for local residents and to alleviate pressure on barren sites selected from the imagery. Each non-bar-

ren site consisted of a 25 m 25 m plot located in a

local ecosystems by replacing fuel wood with electricity

(Li et al. 2001, 2002). During construction of the homogeneous area with a single cover type, and its

hydropower stations, large amounts of wood were cut location was recorded using a GPS with di erential

locally, causing damage to the surrounding ecosystem. correction. Another set of 119 sites (20 coniferous

The disturbance ceased, however, once the construction forest, 21 mixed forest, 20 broadleaf forest, 24 shrub,

was nished, and the ecosystems started to recover 19 herbaceous, and 15 barren sites) was selected in the

through natural succession. According to studies by Liu vicinity of the road network, and assessed in the eld

et al. (1999, 2003), annual fuelwood consumption has for evaluating the accuracy of the classi cation. The

continued to increase from approximately 4,000 to overall accuracy of the classi cation based on these

10,000 m3, causing panda habitat reduction by more evaluation sites was 93.3%. All producer s accuracy

(measure of omission error) and user s accuracy (mea-

than 20,000 ha during the past two decades. A more

sure of commission error) for individual cover types

detailed study on the impacts of three main types of

were >85%. A 3 3 (pixel) four-direction lter proce-

human disturbances will further improve our under-

dure was used to reduce small-scale noise in classi -

standing on human-induced disturbances in panda

cation. The TM imagery had some areas (about 20%

habitat, and thus better protect one of the most endan-

of the reserve area) that were covered with clouds, and

gered species in the world.

thus were excluded from the nal classi cation map.

Since cloud-shadowed areas were mainly near the tops

of mountain ridges, their exclusion had little e ect on

Data sources

the analysis of this study. Most human disturbances

occur in valley areas at relatively low to medium alti-

For our case study, vegetation data for the Wolong

tudes. The digital elevation model (DEM) we used for

Nature Reserve was developed based on TM imagery

spatial analysis in this paper was created from

acquired in 1997. The images were recti ed and

1:100,000 topographic maps by the Chinese Academy

georeferenced using 12 ground control points

of Sciences in 1999 (Ouyang et al. 2000).

(RMS=0.0741). A maximum likelihood supervised

490

Analysis This landscape dissimilarity measure is based only on

landscape composition, although substantial alterations

in landscape composition almost certainly imply altera-

Scope of human disturbances

tions in the spatial structure of such landscapes. Given

GIS-based bu er-zone analysis (Arono 1989) was em- the dimensions of the bu er zones and the resolution of

ployed to evaluate the spatial extent of the major human the data, the use of patch-based landscape metrics to

disturbances on landscape structure and their pattern as quantify landscape structure would not be appropriate

a function of distance from the sources human activi- for this study. An investigation on the landscape pattern

ties associated with road networks, human settlements, and its dynamics in the Wolong Nature Reserve was

and hydropower stations. Linear belt-like bu er zones presented in a related study (Zeng and Wu 2004).

were created along the roads, while concentric rings were

generated around hydropower stations and human set-

Comparison of landscape structure in disturbed versus

tlements (both treated as points). Up to ten consecutive

undisturbed areas

200 m bu er zones were used in the analysis, and eld

observations showed that impacts of these human dis-

In order to understand the nature of the impacts of the

turbances diminished well within 2,000 m of the sources

three types of human disturbance and their relative

of disturbances. Within 2,000 m of disturbance sources,

intensity, we need to compare landscape composition of

altitude can increase up to 400 m in these mountainous

impacted areas to that of relatively undisturbed areas.

terrains. Landscape composition, therefore, can vary

Given the complex pattern of landscape in the mountain

gradually, even in undisturbed areas, along the vertical

terrain, however, comparisons must be made based on

gradient in bu er zones, progressively away from the

the same physical settings that determine the distribu-

valley bottoms. Impact of disturbances, with likely de-

tion of di erent vegetation types in a landscape. To

cay with increasing distance from the source of distur-

enable such comparisons, a database was developed to

bance, can modify the pattern of spatial variation in

classify all the pixels in the study area by speci c phys-

landscape composition along the series of bu er zones.

ical settings. The settings were de ned by combinations

The modi ed patterns may show thresholds that suggest

of the three main physical factors, elevation, aspect, and

the scopes of each type of human disturbances.

slope, derived from the DEM. Elevation (ranging from

The Bray Curtis dissimilarity index, a widely used

1,200 to 6,200 m) was grouped into 25 classes of 200 m

dissimilarity measure in community ecology (Faith

intervals. Five classes of aspects based on exposure

et al. 1987; Legendre and Legendre 1998; Krebs 1999),

(135 225 ; 90 135 and 225 270 ; 45 90 and 270 315 ;

was used to quantify the compositional dissimilarity

0 45 and 315 360 ; and at area) and eight slope

between two landscapes (e.g., two adjacent bu er

classes (0 10, 10 20, 20 30, 30 40, 40 50, 50 60,

zones), referred to as landscape dissimilarity in the

60 70, and >70 ) were used. The comparisons between

remainder of the paper, based on relative abundance of

disturbed and undisturbed areas were based on each of

landscape elements. This dissimilarity index, range

these physical settings represented in a bu er zone of the

from 0 (similar) to 1 (dissimilar), is a standardized

disturbed area. The comparisons were made using

Manhattan distance (d):

landscape dissimilarity between the disturbed and

P

n undisturbed areas based on relative abundance in cor-

jxAi xBi j responding vegetation types (the landscape elements) in

i 1

dAB the two areas. For example, there may be several phys-

Pn

xAi xBi ical settings represented in the 0 200 m bu er zone of

i 1 the settlements. For each of these physical settings,

landscape dissimilarity was calculated for all the pixels

where xAi and xBi are the abundance values of landscape

of a speci c physical setting within the bu er zone

element i in landscapes (e.g., bu er zones) A and B,

against all the pixels of the same physical setting in the

respectively, and n is the number of landscape elements

undisturbed areas. The greater the landscape dissimi-

(vegetation types) in the two landscapes. Obviously, the

larity, the greater the impact of the disturbance in areas

greater the landscape dissimilarity for two adjacent

of the particular physical setting in that bu er zone.

bu er zones, the greater the di erence in landscape

Undisturbed areas were de ned as areas beyond the

structure between the two bu er zones. A curve of the

threshold distance from the sources of human distur-

landscape dissimilarity plotted against the distance from

bance.

a source of human disturbance would show variations in

landscape structure as a function of distance from the

source of human disturbance. We hypothesized a dis- Impact of di erent human disturbances on landscape

tance decay e ect, expecting that the value of the structure

landscape dissimilarity should decrease as distance in-

creases; and furthermore, there should be a threshold The impact of di erent types of human disturbances

distance in the decline of landscape dissimilarity indi- on landscape structure in various bu er zones can be

cating the extent of the impact of human disturbances. assessed based on direct comparisons between the

491

landscape structures of each type of disturbed areas and

corresponding undisturbed area. These comparisons,

however, need to be done by speci c physical settings to

be valid for reasons stated above. Given the large

number of physical settings, these results were aggre-

gated by each bu er zone of each disturbance type to

a ord a synthesized description and comparison of the

patterns of impacts of di erent human disturbances on

landscape structure. For each physical setting in a given

bu er zone of a disturbance type, the deviation in the

percent cover of each vegetation cover type from that of

corresponding undisturbed areas was rst determined.

Average values of these deviations, weighted by the

areas of individual physical settings in the bu er zone,

were then used to describe the general pattern of land-

scape alteration in speci c bu er zones of each distur-

bance type.

Relative importance of di erent human disturbances

Fig. 2 Landscape dissimilarities between adjacent (200 m) bu er

zones of each disturbance types. The landscape dissimilarity was

The areas impacted by di erent sources of human dis-

measured using the Bray Curtis dissimilarity index based on

turbances often overlap in Wolong Natural Reserve. abundance values (in area) of the six vegetation cover types

Comparisons between di erent overlapping areas in (landscape elements): coniferous forest, mixed coniferous broad-

their landscape dissimilarity to undisturbed areas can leaf forest, broadleaf forest, shrub, herbaceous, and barren land

provide useful information on the relative importance of

di erent types of human disturbances. There were three

a certain distance from the source of human disturbance,

types of overlapping disturbed areas: areas impacted by

landscape structure is mainly a ected by natural factors,

both roads and settlements, areas impacted by both

and the impact of human activities was insigni cant.

roads and hydropower stations, and areas impacted by

Landscape dissimilarity values for all three human im-

roads, settlements, and hydropower stations. Some of

pacts became stable around the 1,000 1,200 m bu er

the disturbed areas were impacted solely by roads or

zones, which can be regarded as the threshold distance

settlements, but all areas impacted by hydropower sta-

within which human activities had substantial impacts

tion were also impacted by roads. Each pair of these ve

on the surrounding landscape. The only subtle di erence

types of disturbed areas was compared using a paired

was that settlements appeared to extend their in uence

t-test between corresponding landscape dissimilarity (to

slightly further (up to 1,200 m) than did roads and

undisturbed areas) values of the two types by bu er

hydropower stations (always within 1,000 m).

distance classes (up to 1,000 m) and physical settings.

Near the sources of the disturbances (0 200 and 200

For example, when comparing areas impacted only by

400 m bu er zones), landscape dissimilarity has the

roads and areas impacted by both roads and settlements,

greatest values for hydropower stations, medium for

if the impacted areas were within the rst ve 200 m

settlements, and the least for roads. These results indi-

bu er zones and each bu er zone had 60 common

cate that construction of hydropower stations resulted in

physical settings, a t-test of 300 pairs of landscape

greater impacts on the immediately surrounding land-

dissimilarity values would be conducted. A signi cance

scape than did human activities associated with settle-

level of 0.05 was used for all t-tests.

ments and roads. It should be noted that although the

impact of settlements on immediately surrounding

Results landscape is at the medium level, its threshold distance

for appreciable alterations in landscape structure is the

greatest among all three sources, re ecting that human

As re ected clearly from the analysis of the landscape

activities around residential settlements have more

dissimilarity of adjacent bu er zones around the three

extensive impacts.

sources of human disturbance (Fig. 2), human activities

The frequency distributions of the types of disturbed

exerted substantial impacts on the landscape structure.

areas (within a 1,000 m bu er zone of sources of dis-

Intensity of the impacts tended to decline as distance

turbance) showed the altitudinal distribution of the

from the source of the human disturbance increases.

areas impacted by hydropower stations ($1,400

When the distance reached a threshold, landscape dis-

3,200 m) was slightly lower than that of the areas im-

similarity between adjacent bu er zones tended to sta-

pacted by settlements ($1,800 3,800 m), while the areas

bilize at low levels with small uctuations, suggesting

impacted by roads had a much wider range ($1,600

that landscape structure varied only minimally from one

5,000 m) (Fig. 3a c). The proportion of non-forested

bu er zone to the next. This demonstrated that, beyond

492

A direct result of human impacts is that forest cover has

decreased substantially, while the proportion of shrub

and barren lands, which are indicators of ecological

degradation (especially at lower elevations) has in-

creased remarkably. The results in Fig. 4 also showed

that the intensity of the impact decreased with the

increasing distance from the sources of disturbance.

The impact of human activities in Wolong Nature

Reserve, a typical alpine canyon area, appeared to be

con ned in fairly restricted areas. Landscape structure in

the rst three bu er zones was altered signi cantly, with

remarkable decreases in forest cover coupled by in-

creases in non-forest cover. From the fourth bu er zone

outwards, alteration of landscape structure becomes less

signi cant. This was consistent with the results on the

scope of human impact on landscape (Fig. 2) and also

suggested that the most intensive human impacts oc-

curred within the 600 m radius from the sources of

disturbance, the 600 1,000 m bu er zone appeared to be

a transitional area between human-impacted and

undisturbed landscapes, and the areas beyond 1,000 m

were dominated by undisturbed landscape.

The results in Fig. 4 showed that hydropower sta-

tions and settlements were the dominant foci of human

impact on landscapes in the reserve, consistently with

the results on the scope of human impact on landscape

(Fig. 2). In contrast, the impact of human activities

along roads was less signi cant. Forest cover decreased

the most in bu er zones of hydropower stations ($40%

in the 200 m bu er zone), and the percent cover of

barren areas was also highest in these areas (Fig. 4a).

Fig. 3 Altitudinal distribution of areas impacted by di erent types

The impact around settlements, measured by overall

of human disturbances, as well as the relative abundance of

alteration of vegetation covers within the bu er zones,

vegetation cover types at each elevation. The six vegetation cover

types were: coniferous forest (Conifer), mixed coniferous broadleaf was lower than that of hydropower stations (Fig. 4a,b).

forest (Mixed), broadleaf forest (Broadleaf), shrub, herbaceous,

The conifer forest component of the altered forest cover

and barren land (Barren)

tended to increase in bu er zones that are farther away

from the source of disturbance, probably also related to

increase in elevation in increasingly distant bu er zones.

The impact along roads (Fig. 4c) was far less signi cant

cover in bu er zones of both hydropower stations and

than those around hydropower stations and settlements,

settlements was higher than that in bu er zones of roads

which may be partially attributable to the implementa-

at lower elevations (

Contact this candidate